Daily blood pressure profile and blood–brain barrier permeability in patients with cerebral small vessel disease

Cerebral small vessel disease (CSVD) plays an important role in cognitive impairment, stroke, disability, and death. Hypertension is the main risk factor for CSVD. The use of antihypertensive therapy has not resulted in the expected decrease in CSVD complications, which may be related to the underestimation of significance of daily blood pressure profile for blood–brain barrier (BBB) permeability. 53 patients with CSVD of varying severity (mean age 60.08 ± 6.8 years, 69.8% women, subjects with treated long-standing hypertension vs. normotensive subjects − 84.8% vs. 15.2%) and 17 healthy volunteers underwent ambulatory blood pressure monitoring (ABPM) and MRI, including T1-weighted dynamic contrast-enhanced magnetic resonance imaging for assessing BBB permeability. Most of ABPM parameters in CSVD patients did not differ from controls, but were associated with the severity of white matter hyperintensity (WMH) and the total CSVD score. BBB permeability in normal-appearing white matter (NAWM) and grey matter (GM) was significantly higher in CSVD patients, and the severity of BBB permeability remained similar in patients with different stages of WMH. Among BBB permeability parameters, the area under the curve, corresponding to an increase in the contrast transit time in NAWM, had the greatest number of correlations with deviations of ABPM parameters. BBB permeability in CSVD is a universal mechanism of NAWM and GM damage associated with a slight increase in ABPM parameters. It is obvious that the treatment of hypertension in patients with not severe WMH should be more aggressive and carried out under the control of ABPM.

www.nature.com/scientificreports/ Hypertension is the main risk factor for age-related CSVD [4][5][6] . In most cases, the severity of hypertension correlates with MRI markers of CSVD, such as white matter hyperintensity (WMH) and lacunae 4,7 . The leading mechanism of brain damage in patients with chronic hypertension is hypoperfusion secondary to arteriolosclerosis, which is characterized by the loss of smooth muscle cells, accumulation of fibrotic hyaline deposits, thickening of the blood vessel walls, and luminal narrowing 8,9 . Modern antihypertensive therapy, aimed primarily at preventing remodelling of small arteries and, accordingly, increasing blood flow, has led to a decrease in the incidence of stroke 10,11 , but not the prevalence of cognitive impairment 12,13 . This can partly be explained by a lack of or a faint effect of antihypertensive therapy on the increased blood-brain barrier (BBB) permeability 14 , which, along with ischemia, is another significant mechanism of brain damage in hypertension [15][16][17] . BBB damage in patients with hypertension is considered to be the main mechanism for the initiation and progression of CSVD and the development of forms mixed with neurodegeneration [15][16][17] . In these cases, hypertension is both a factor in BBB damage 18 and a consequence of the damage to the cerebral autonomic centres caused by the high BBB permeability 19 .
For a long time, BBB damage with high permeability has been considered mainly as a failure of cerebral autoregulation due to high blood pressure in acute and chronic hypertension 20,21 . Further experiments on the spontaneously hypertensive rats and stroke-prone spontaneously hypertensive rats have proven that the mechanism of BBB damage in CSVD is universal and can be observed in milder hypertension stages 22 . One of the possible explanations may be the effect of blood pressure variability on the high BBB permeability. This assumption is consistent with recent studies that have noted the importance of blood pressure variability in the development of CSVD, including in controlled hypertension according to outpatient measurements 23,24 .
In vivo study of the role of this mechanism in CSVD development has become possible with T1-weighted dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), enabling a quantitative assessment of BBB permeability 25,26 . Increased BBB permeability was found in normal-appearing white matter (NAWM) as compared to controls 27,28 ; in WMH and adjacent NAWM, correlating with the severity of WMH, hypertension, age and leading to the delayed reduction in cognitive capacity 25 .
Thus, hypertension and increased BBB permeability are of great importance in the development of CSVD 17,25,27,28 , but antihypertensive therapy has insufficient effect on reduction of cognitive impairment 12,13 . Of particular interest is the clarification of the relationship between the features of hypertension, which is well controlled by outpatient measurements, and BBB permeability in patients with CSVD and cognitive impairment. The study aims to evaluate the relation between daily blood pressure profile and BBB permeability in patients with CSVD and cognitive impairment.

Materials and methods
Participants and ethics. The study included patients aged 46-70 years with cognitive and other neurological complaints, such as gait and balance problems, mood disorders or residual symptoms after stroke, brain changes on MRI corresponded to CSVD (WMH, lacunae, enlarged perivascular spaces, microbleeds and cerebral atrophy) 29 . Patients with low WMH burden (Fazekas scale score 1) were included in the study if they had hypertension stage 2 or 3 and/or ≥ 1 lacuna.
The control group consisted of volunteers with no clinical or MRI evidence of vascular and degenerative brain pathology, no hypertension in the medical history and during Ambulatory Blood Pressure Monitoring (ABPM), and matched for age and gender. Controls with hypertension according to ABPM were excluded from the study, in accordance with the European Society of Hypertension recommendations: awake blood pressure was ≥ 135/85 mmHg and/or asleep blood pressure was ≥ 120/70 mmHg, or if blood pressure increased by more than 24% over time during exertion 32 .
In total 53 patients (37 women, average age 60.1 ± 6.8 years) and 17 healthy volunteers (12 women, average age 56.7 ± 6.7 years) were enrolled in the study.
Traditional vascular risk factors, such as hypertension 33 , hypercholesterolemia, obesity, diabetes mellitus and smoking were assessed in the patients and controls.
The study was approved by the Local Ethics Committee of the Research Centre of Neurology № №2-4/16 dated 17.02.2016 and performed in accordance with the principles of the Declaration of Helsinki. All subjects signed an informed consent form for participation in the study.
Ambulatory blood pressure assessment. All participants underwent ABPM with an automated device (LLC DMS Advanced Technologies, Moscow) based on oscillometric method. Patients underwent ABPM during hospitalization with blood pressure measurement every 30 min during the day (8:00 am to 10:00 pm) and every 60 min during the night (10:00 pm to 8:00 am). The ABPM device inflatable cuff was placed on the nondominant upper limb. In all cases, at least 70% of the measurements were suitable for analysis. We calculated mean 24-h systolic blood pressure (SBP) and diastolic blood pressure (DBP); mean, standard deviation and maximal values of awake and asleep SBP and DBP; and blood pressure load parameters as the percentage of readings in a given period (24-h, day, or night), which exceeded the normal levels for awake and asleep SBP and DBP 32 . www.nature.com/scientificreports/ The grade of hypertension was determined from the medical history and was adjusted according to ABPM results. During hospitalization patients continued their antihypertensive therapy.
Two neuroradiologists evaluated MR images in a standardized manner, blinded to clinical information. No STRIVE criteria were found in volunteers from control group. There were no acute or recent small lacunar infarcts based on DWI analysis in patients with CSVD. MRI presence of lacunes, white matter hyperintensities, microbleeds, and perivascular spaces were summed in a score of 0-4 representing all CSVD features combined 34,35 .
Image processing. The entire DCE-MRI dataset underwent preliminary processing using the NordicNeu-roLab software (NordicICE, Norway). This included automatic correction of motion artefacts, correction of preand post-contrast data in the dynamic series, concentration of contrast agent in the brain tissue calculation using relative signal change and T1 mapping. Individual vascular input functions were derived semi-automatically from the superior sagittal sinus 37 . The hematocrit, contrast agent dose and relaxivity of the contrast agent were set individually for each patient. The Patlak pharmacokinetic model was used to assess the low BBB permeability in CSVD resulting in Ktrans (volume transfer coefficient), Vp (fractional blood plasma volume) maps, and AUC (area under the curve-corresponding to increased contrast transit time in the brain) maps (Fig. 1A).
Once permeability parameter maps were obtained, further data processing was performed in SPM12 (http:// www. fil. ion. ucl. ac. uk/ spm). This included the following steps: coregistration of each subject's permeability parameter maps and the T1-weighted images; segmenting the T1-weighted images into grey matter and white matter, followed by the correction of obtained images using WMH masks based on a MatLab script (https:// matlab. ru/), resulting in the binary images of the corrected grey and white matter. Permeability parameters were calculated in ITK-SNAP (http:// itksn ap. org) separately for grey matter (GM), NAWM and WMH by superimposing the relevant masks over the individual permeability maps (Fig. 1B). BBB permeability parameters have very low values (10 −4 ), which do not visually differ, therefore, an analysis of the data of a patient with CSVD stage 2 according to Fazekas Scale is given as an example. software. Data are presented as n (%) for categorical variables or as mean ± standard deviation (SD) or median [interquartile range (IQR)] for quantitative data. Differences between groups were determined using χ2, independent samples t-test or Mann-Whitney test, univariate analysis of variance or Kruskal-Wallis test with Bonferroni correction, where appropriate. In all cases, two-way statistical criteria were used. The null hypothesis was rejected if p < 0.05. Pearson's correlation coefficient and Spearman's correlation were used to assess the relationship between parameters.

Results
CSVD and the control groups were matched for age and gender, and consisted predominantly of women (Table 1). Vascular risk factors were comparable except for hypertension, which was the dropout criterion for the control group.
Most of the patients in the main group (84.8%) had hypertension of varying severity and were taking one or more antihypertensive drugs.
The main disease symptoms were cognitive impairment, gait disturbances unrelated to post-stroke hemiparesis, and MRI changes including WMH, lacunae, microbleeds and dilated perivascular spaces (Table 2).
Differences in mean asleep DBP, and asleep SBP and DBP load were found when ABPM results were compared between subjects with CSVD and those in the control group (Table 3).
ABPM results showed a good response to antihypertensive therapy in the main group. Most of ABPM parameters in the study (except for mean asleep DBP, asleep SBP load, and asleep DBP load) did not show intergroup www.nature.com/scientificreports/ differences corresponding to doctors' and patients' opinion that hypertension was well controlled as measured on the outpatient basis. Although, there were found the relations of ABPM results with neuroimaging markers of CSVD (Table 4). Increased SBP and DBP values had significant relations with WMH load, based on Fazekas Scale score, and correlations with WMH volume and total CSVD score.  www.nature.com/scientificreports/ Relationships between the standard deviation of blood pressure, which correspond to the indices of variability, and neuroimaging markers of CSVD were not found either by comparative or correlation analysis, therefore these data are not included in the table.
To clarify the link between daily blood pressure fluctuations and BBB permeability, the last one was assessed using DCE-MRI in patients and controls (Fig. 2).
According to DCE-MRI, all the study parameters of BBB permeability in NAWM and GM, except for Ktrans in NAWM, were significantly higher in patients with CSVD than in controls.
BBB permeability decreased as WMH Fazekas score increased, with significant differences in Vp and AUC in WMH (Table 5).
Differences in the BBB permeability parameters in GM and NAWM depending on the WMH Fazekas score were not found.
Statistically significant correlations were seen between AUC in GM and NAWM, Vp in NAWM and the parameters of 24-h and awake SBP and DBP (Table 6).   www.nature.com/scientificreports/ Increased BBB permeability, as assessed by AUC, was correlated with mean awake DBP, maximal awake SBP and DBP, awake SBP load (%) in GM, as well as with mean awake DBP and asleep SBP, maximal awake and asleep SBP and DBP, awake SBP load (%) in the NAWM.
AUC, which characterizes the contrast delay in the brain, had the highest sensitivity out of BBB permeability parameters.
No significant correlations were found between ABPM results and BBB permeability parameters in WMH, so these data are not provided.

Discussion
This study sought to clarify the relation between treated hypertension characteristics, assessed by ABPM, and BBB permeability, assessed by DCE-MRI, in patients with CSVD. Most of ABPM parameters in the study did not show intergroup differences, which confirms a good response to antihypertensive therapy in the main group. However, comparison of ABPM results with the severity of WMH based on Fazekas Score and its volume, as well as with total CSVD score, showed direct and significant relations. These data indicate the presence of certain mechanisms related to the abnormal ABPM parameters in patients compared with controls. Since a significant proportion of the studied patients had mild hypertension, as well as mild clinical and MRI signs of the disease, we could assume that BBB damage and its high permeability played a significant role in CSVD development. This hypothesis is based on the study results that indicate the special role of endothelial dysfunction with high BBB permeability as a mechanism of early CSVD 25,38-40 . According to DCE-MRI, all the study parameters of BBB permeability in GM and NAWM, except for Ktrans in NAWM, were significantly higher in patients with CSVD compared to controls that is consistent with the previous studies [26][27][28]41 . There were no significant differences in BBB permeability in GM and NAWM between patients with WMH of varying severity. It should be noted that the results of previous studies regarding the significance of BBB permeability in the development of early or late CSVD are contradictory 25,28,41 . A reasonable explanation for our data may be that an increase in BBB permeability is possible with a relatively intact endothelium of small vessels in GM and NAWM in CSVD, corresponding the data about the universal mechanism of BBB damage in patients with CSVD and hypertension of varying severity 22 . On the other hand, the obtained data on a decrease in BBB permeability in the WMH may represent the conditions that characterize late-stage CSVD such as progressive endothelial death, impaired autoregulation due to small vessel remodelling, wall thickening and lumen narrowing, reduced microvascular perfusion 9,28 . This explanation is also supported by the fact that none of the abnormal ABPM parameters had relations with increased BBB permeability in the WMH. This result matches the different responses to elevated blood pressure in normotensive and hypertensive rats in following experiment. The hypertensive rats had higher permeability to sucrose which was absorbed more slowly by the www.nature.com/scientificreports/ brain, and the authors attributed this to changes in blood flow in hypertension 18,42 . AUC, which characterizes the contrast delay in the brain, had the highest sensitivity out of BBB permeability parameters. In our study, AUC in GM and NAWM had the most of relations with ABPM parameters. Although the latter in patients with CSVD did not differ from ones of the control group, we cannot exclude preceding rises in blood pressure that exceeds the upper threshold of cerebral autoregulation with an increase in BBB permeability 21 . It can be assumed that the use of antihypertensive therapy may change the upper threshold of cerebral autoregulation and the conditions for its disruption. This statement is indirectly supported by the fact that a greater reduction in blood pressure in elderly people with hypertension is associated with increased cerebral blood flow, corresponding to a shift in the autoregulation curve 43 . On the whole, these data support the necessity for more aggressive treatment of hypertension in patients with CSVD 44,45 . The risk of cardiovascular complications is decreased when blood pressure is reduced more aggressively, so the guidelines were rationally revised to the target SBP < 120 mmHg 46 . A recent randomized trial also supported this finding, as cerebral perfusion did not decrease in patients with severe CSVD when blood pressure was aggressively reduced, unlike in healthy controls 47 .
The obtained data on the universal nature of increased BBB permeability in the NAWM and GM in patients with CSVD indicate an ongoing pathological process in the small arteries, which leads to endothelial damage in relatively well-preserved small vessels. The connection between BBB permeability in NAWM and GM and elevated ABPM parameters indicated the importance of autoregulation dysfunction in promoting this mechanism. It is possible that the underestimation of the pathological mechanism of brain damage due to increased BBB permeability can partly explain the significance of hypertension in middle age for the development of cognitive impairment in the elderly 48,49 .
The lack of significant differences in ABPM results related to BBB permeability between patients and controls allow us to hypothesize the presence of additional factors of endothelial damage and increased BBB permeability alongside hypertension. These factors may be chronic inflammation 50 or high salt sensitivity, which have been found to independently correlate with CSVD 51 .
The results of the study point out the necessity for more aggressive treatment of hypertension and repeat usage of ABPM as well as requirement for searching and affecting factors that potentiate the role of hypertension in CSVD development. It is obvious that further studies are needed on the effect of antihypertensive therapy on BBB permeability and its ability to protect the brain from damage in patients with CSVD.

Data availability
Raw data were generated at Research Center of Neurology. The data that support the findings of this study are available from the corresponding author upon reasonable request. Clinical, neurovisualization and statistical data will be available upon request from any qualified investigator.